Your Generator Hasn’t Really Run Until a Load Bank Made It Sweat

A customer once asked me to sit in on his generator’s monthly test. The thing started on its own, ran for twenty minutes humming along at no load, and shut itself down. He turned to me and smiled: “Well, at least that part works.” He was right about exactly one thing — that part worked. Everything the machine would be asked to do in a real outage? None of that had ever been tested. Not once, in six years.

That gap is what load bank testing exists to close. And it’s the subject of three questions I get more than any others. Let me answer them the way I answer them over coffee.

Question 1: “My set starts every month by itself. Isn’t that enough?”

Honestly? That monthly auto-start is the generator equivalent of walking to the fridge and calling it cardio.

Here’s what actually happens in a no-load run: the engine fires up, idles at 1500 or 1800 RPM with nothing drawing power, and shuts down before anything inside it gets properly hot. The cylinders never reach the temperatures the engine was designed around. Unburned diesel starts glazing the cylinder walls — a condition called wet stacking — and oil and fuel residue creep out through the exhaust. I’ve seen exhaust stacks on “perfectly healthy” standby sets that wept black, wet soot you could smear with a finger. That’s years of no-load exercise written in carbon.

This isn’t just my opinion. The US fire-safety standard NFPA 110 lays it down plainly: if your monthly run can’t hold at least 30% of the nameplate rating, you’re supposed to do a supplemental load test that does. Thirty percent. Most standby sets I walk up to never see half of that in their entire working lives, because they’re oversized for their actual emergency loads. A 400 kVA set feeding 60 kW of emergency lighting and one elevator is the classic case — the owner bought a racehorse and makes it walk laps around the paddock.

So no. Starting isn’t enough. Starting is the easy part. The hard part is working.

Question 2: “Fine. So what actually happens during the two hours?”

A load bank is a big steel box of resistive heating elements that pretends to be your building. A tech cables it to the generator’s output terminals, starts the set, and starts dialing in load in steps — typically 25, 50, 75, and 100 percent of the rated kW. Two hours is the standard appointment. Every Pavo set that leaves our factory gets exactly that before it ships: two hours on the load bank, full rating, with the factory’s own test engineers signing off.

During those two hours, the tech watches five things:

• kW and kVA output — does it actually make its nameplate numbers?
• Frequency — does it hold steady when load steps on, or does it sag and wander?
• Voltage — same question, different meter.
• Coolant temperature — a set that runs cool at no load can boil at full load.
• The exhaust — color, smoke, and whether the stack is weeping wet fuel.

And this is the part nobody tells you beforehand: a shocking number of sets fail something. I’ve watched a “fine” standby generator trip at 85 percent of its rated load — a governor problem nobody had ever found, because nobody had ever asked the machine to do 85 percent of anything. In a real blackout, that set would have run for about eleven minutes and died. The load bank found it on a Tuesday afternoon instead of at 2 a.m. in a storm. That’s the whole point.

For prime-power sets, the test can go further — 110 percent of rating for an hour. That’s allowed by ISO 8528 (one hour of 10 percent overload in every twelve hours of running), and it’s how you find out whether the cooling system has any real margin. A step load rejection at that level — suddenly dropping a big chunk of load — is also where you learn whether the voltage regulator and governor were sized properly or just look it on paper.

Question 3: “Honest answer — is this a real test, or a line item on an invoice?”

Fair question. And I’ll give you an answer that costs me agreement with some of my own colleagues: not every generator needs one.

A prime-power set running a mine at 70 percent load, ten hours a day? The job is the test. Paying someone to bring a load bank to that machine is theater. But an oversized standby set that serves a hospital wing’s emergency circuit and sees no load for months at a stretch? Once a year — it belongs on the same annual schedule as your full maintenance service. No argument from me.

There’s a middle case I get asked about a lot: smaller home standby sets. The NFPA 110 rule technically targets bigger installations, so your 20 kVA home unit isn’t failing any law by skipping it. But if you’ve ever noticed wet, oily deposits at the exhaust tip — the weeping I described earlier — that machine has been light-loaded too long, and a load bank session will burn that buildup off better than anything else you can do to it.

One more thing buyers hear and get confused by: resistive versus reactive load banks. A resistive bank is unity power factor — pure heat, pure kW. A reactive bank adds inductive load to simulate motors, testing at 0.8 power factor. For most owners, a resistive test answers every question that matters. The reactive version is for data centers and hospitals that want to see how the alternator handles motor-starting current. If a test company quotes you extra for the reactive version, ask what load mix you actually have before you say yes.

So is it a real test? Here’s my test for whether the test is real: ask your tech what percentage of the nameplate rating he loaded your set to last time. If he answers with a number — “we took it to 100 percent for two hours” — you bought a real test. If he goes quiet, or says “we ran it up real good,” you bought a line item.

Generators don’t lie about what they can do. They just never get asked. Until the night it matters.